Tuesday, February 6, 2007

Study on Wikipedia accuracy in History

It might be best to let Steven Colbert have the last word on Wikipedia, but since others are talking about it (skeptics/critics Brian Leiter and Dan Solove; Michael Dorf, defending; Jeff Lipshaw occupying something of a middle ground), I have another offering from the history world. Roy Rozenswieg, history professor at George Mason and Director of the Center for History and New Media may not be a household name in the law world, but he is a historian who has devoted much of his career to bringing history, and historians, to the web. Not known as a Luddite, in other words. He evaluated Wikipedia for the Journal of American History, the premier American history journal, and the results are here.

Wikipedia has many advantages, but how does its accuracy hold up in the field of history? Here's Rozenswieg:

American historians might look first at the Wikipedia page headed “List of United States History Articles,” which includes twelve articles surveying American history in conventional time periods and another thirty or so articles on such key topics as immigration, diplomatic history, and women’s history. Unfortunately, the blind man reporting from those nether regions would return shaking his head in annoyance. He might start by complaining that the essay on the United States from 1918 to 1945 inaccurately describes the National Industrial Recovery Act of 1933 as in part a response to the “dissident challenges” of Huey Long and Father Charles Coughlin—a curious characterization of a law enacted when Coughlin was still an enthusiastic backer of Roosevelt and Long was an official (if increasingly critical) ally. But he would be much more distressed by the essay’s incomplete, almost capricious, coverage than by the minor errors. Dozens of standard topics—the Red Scare, the Ku Klux Klan, the Harlem Renaissance, woman suffrage, the rise of radio, the emergence of industrial unionism—go unmentioned. And he would grind his teeth over the awkward prose and slack analysis (“the mood of the nation rejected Wilson’s brand of internationalism”) and the sometimes confusing structure (the paragraph on legislation passed in 1935 appears in the section on Roosevelt’s second term).

Other entries in the United States history series are worse. The entry on women leaves out the Nineteenth Amendment but devotes a paragraph to splits in the National Organization for Women (now) over the defense of Valerie Solanas (who shot Andy Warhol).

He found that Wikipedia tended to be better in biographical entries, however. Compared to other on-line encyclopedias, "Wikipedia, then, beats Encarta but not American National Biography Online in coverage and roughly matches Encarta in accuracy."

And this ultimately leads to a point that has been obscured in some posts. Wikipedia, like other encyclopedias, has its place. But when should judges, law professors and students be citing to encyclopedias, on-line or otherwise? Rozenswieg quotes one academic as calling encyclopedias "the Reader’s Digest of deep knowledge."

How about this for a ground rule: if the authority you need in that footnote in your opinion or article is on the level of Reader's Digest, go ahead and cite to Wikipedia.

UPDATE: For a different take on the topic, Mike Madison focuses on Wikipedia governance, linking to a Harvard Business School case study based on Wikipedia.